My research concerns philosophical issues in scientific modeling and experimentation, and emphasizes the transformative effect that new technologies have on these practices. General questions of interest include: What makes theoretical abstractions informative with respect to natural phenomena? How do we judge the reliability of our means of generating and processing data? How are such strategies affected by technological change, especially when these become automated and result in massive amounts of data? And how do these effects condition scientists’ conception of their target of inquiry, including socially salient categories like race?
To address these, I focus on the related fields of molecular, systems, and computational biology, which have recently experienced a rise of new modeling and data-gathering practices pertinent to these issues. I am developing a framework that can address them from an integrated philosophical and historical standpoint. It proceeds along three strands: (1) understanding how models are inferentially connected to data-gathering practices in science; (2) clarifying the epistemic norms and frameworks guiding scientific methodology and explanation; (3) investigating the historical interplay of experimental technologies and conceptions of natural phenomena.
These interests have led me to study a broad range of topics, from recent uses of AI in biology to early modern thinking on the role of experiment and hypotheses.
Publications
“Understanding Phenomena Through Opaque Models: Machine Learning and Integrative Practices in Structural Biology” in D. Rowbottom, A. Curtis-Trudel, and D. Barack (eds), The Role of AI in Science: Epistemological and Methodological Studies (forthcoming)
“Representational Content in Minds and Models: Latent Influences and Explanatory Challenges” in Z. Kondor and I. Danka (eds), Representation in Ab/Use (forthcoming)
“Crystallizing Techniques: Sample Preparations, Technical Knowledge, and the Characterization of Blood Crystals, 1840-1909”, History and Philosophy of the Life Sciences (forthcoming, preprint)
“Observations, Experiments, and Arguments for Epistemic Superiority in Scientific Methodology” (with Nora Boyd), Philosophy of Science, 91(1): 111-131 (preprint)
“Empirical Techniques and the Accuracy of Scientific Representations,” Studies in the History and Philosophy of Science, Part A, 94: 143-157 (preprint)
“The Rise of Cryptographic Metaphors in Boyle and Their Use for the Mechanical Philosophy,” Studies in the History and Philosophy of Science, Part A, 73: 8-21 (pdf)
“Mechanistic Explanation in Systems Biology: Cellular Networks,” British Journal for the Philosophy of Science, 68(1): 1-25 (pdf)
“Causal Concepts Guiding Model Specification in Systems Biology,” Disputatio, Special Issue on Causality and Modeling in the Sciences, 9(47): 499-527 (pdf)
Sample abstracts
“Understanding Phenomena Through Opaque Models: Machine Learning and Integrative Practices in Structural Biology”
This paper connects structural biologists’ uses of new algorithms for protein structure prediction with philosophical discussion of the epistemology of machine learning. One prominent area of discussion concerns the epistemic opacity, interpretability, and explainability of deep learning models. I argue that structural biologists have developed methods and norms for using AI that, when properly followed, effectively circumvent the most serious obstacles to understanding. I highlight three interrelated reasons for this being the case and use these points to think about the relation between scientific understanding and machine learning algorithms. I propose we view scientific understanding as the outcome of a distributed but coordinated set of scientific activities, rather than something that resides in the internal content of a model.
“Fine-Graining Accounts of Exploratory Experimentation”
Philosophers of science commonly approach both domains in terms of the distinction between (a) hypothesis- or theory-driven experimentation and (b) exploratory experimentation, typically locating these experiments within the latter category. This has led to debates over how theories relate to exploratory experimentation in these contexts. Such debates speak to a lack of specificity in the underlying concept of exploratory experimentation and illustrate the need for a more detailed understanding of the diverse ways that theory is related to experimental practice. Following this lead, I compare the data production and processing pipelines across recent experiments at ATLAS and in high-throughput biology, analyzing the factors that inform choices in each process. Going beyond the distinction between general and local theory, I argue for a “reticulated” account of the relation between theory and experimental practice. The differences between the two cases brought out by this account aid in diagnosing concerns over the reliability of recent methods in molecular biology.
“Pragmatic Representational Content”
I propose a notion of the representational content of scientific models that resists the dialectic over scientific representation inherited from philosophical semantics. This view neither reduces the content of models to cognitive functions, nor appeals to a model-target relation as that which determines a model’s content. Instead, I frame scientific representation as the use of a model within a system of practice. I argue that there are features of this practice, beyond a user’s cognitive acts, that determine the representational content of a model. From this, I articulate a view of pragmatic representational content understood as the set of action- and expectation-guiding inferences contributed by a model when integrated with an empirical practice. Each of the key terms is given further explication, followed by an account of what “grasp” of this content consists in.
My research concerns philosophical issues in scientific modeling and experimentation, and emphasizes the transformative effect that new technologies have on these practices. General questions of interest include: What makes theoretical abstractions informative with respect to natural phenomena? How do we judge the reliability of our means of generating and processing data? How are such strategies affected by technological change, especially when these become automated and result in massive amounts of data? And how do these effects condition scientists’ conception of their target of inquiry, including socially salient categories like race?
To address these, I focus on the related fields of molecular, systems, and computational biology, which have recently experienced a rise of new modeling and data-gathering practices pertinent to these issues. I am developing a framework that can address them from an integrated philosophical and historical standpoint. It proceeds along three strands: (1) understanding how models are inferentially connected to data-gathering practices in science; (2) clarifying the epistemic norms and frameworks guiding scientific methodology and explanation; (3) investigating the historical interplay of experimental technologies and conceptions of natural phenomena. These interests have led me from recent uses of AI in biology to early modern thinking on the role of hypotheses.
These interests have led me to study a broad range of topics, from recent uses of AI in biology to early modern thinking on the role of experiment and hypotheses.
Publications
“Understanding Phenomena Through Opaque Models: Machine Learning and Integrative Practices in Structural Biology” in D. Rowbottom, A. Curtis-Trudel, and D. Barack (eds), The Role of AI in Science: Epistemological and Methodological Studies (forthcoming).
“Representational Content in Minds and Models: Latent Influences and Explanatory Challenges” in Z. Kondor and I. Danka (eds), Representation in Ab/Use (forthcoming).
“Crystallizing Techniques: Sample Preparations, Technical Knowledge, and the Characterization of Blood Crystals, 1840-1909” (forthcoming, History and Philosophy of the Life Sciences).
“Observations, Experiments, and Arguments for Epistemic Superiority in Scientific Methodology” (with Nora Boyd), Philosophy of Science, 91(1): 111-131 (preprint)
“Empirical Techniques and the Accuracy of Scientific Representations,” Studies in the History and Philosophy of Science, Part A, 94: 143-157 (preprint)
“The Rise of Cryptographic Metaphors in Boyle and Their Use for the Mechanical Philosophy,” Studies in the History and Philosophy of Science, Part A, 73: 8-21 (pdf)
“Mechanistic Explanation in Systems Biology: Cellular Networks,” British Journal for the Philosophy of Science, 68(1): 1-25 (pdf)
“Causal Concepts Guiding Model Specification in Systems Biology,” Disputatio, Special Issue on Causality and Modeling in the Sciences, 9(47): 499-527 (pdf)
Sample abstracts
“Understanding Phenomena Through Opaque Models: Machine Learning and Integrative Practices in Structural Biology”
This paper connects structural biologists’ uses of new algorithms for protein structure prediction with philosophical discussion of the epistemology of machine learning. One prominent area of discussion concerns the epistemic opacity, interpretability, and explainability of deep learning models. I argue that structural biologists have developed methods and norms for using AI that, when properly followed, effectively circumvent the most serious obstacles to understanding. I highlight three interrelated reasons for this being the case and use these points to think about the relation between scientific understanding and machine learning algorithms. I propose we view scientific understanding as the outcome of a distributed but coordinated set of scientific activities, rather than something that resides in the internal content of a model.
“Fine-Graining Accounts of Exploratory Experimentation”
Philosophers of science commonly approach both domains in terms of the distinction between (a) hypothesis- or theory-driven experimentation and (b) exploratory experimentation, typically locating these experiments within the latter category. This has led to debates over how theories relate to exploratory experimentation in these contexts. Such debates speak to a lack of specificity in the underlying concept of exploratory experimentation and illustrate the need for a more detailed understanding of the diverse ways that theory is related to experimental practice. Following this lead, I compare the data production and processing pipelines across recent experiments at ATLAS and in high-throughput biology, analyzing the factors that inform choices in each process. Going beyond the distinction between general and local theory, I argue for a “reticulated” account of the relation between theory and experimental practice. The differences between the two cases brought out by this account aid in diagnosing concerns over the reliability of recent methods in molecular biology.
“Pragmatic Representational Content”
I propose a notion of the representational content of scientific models that resists the dialectic over scientific representation inherited from philosophical semantics. This view neither reduces the content of models to cognitive functions, nor appeals to a model-target relation as that which determines a model’s content. Instead, I frame scientific representation as the use of a model within a system of practice. I argue that there are features of this practice, beyond a user’s cognitive acts, that determine the representational content of a model. From this, I articulate a view of pragmatic representational content understood as the set of action- and expectation-guiding inferences contributed by a model when integrated with an empirical practice. Each of the key terms is given further explication, followed by an account of what “grasp” of this content consists in.